Are you in a depression over compression? Have no fear; your audio shrink is here. Compression is an effect that travels under the radar but, especially in sound reinforcement applications, might be the most important tool at your disposal.
Diagnosis, Please
Compression reduces dynamic range by making loud sounds softer. A common misconception is that compression makes soft sounds louder, but that is not the case. Compression provides an illusion that soft sounds are louder, because after we have reduced the level of the peaks, we often need to add gain to bring up the overall level. Hardware compressors employ a variety of technologies for gain control including Voltage Controlled Amplifiers (VCA’s), Field-Effect Transistors (FET’s) or Opto-Isolator circuitry. Many software plug-ins model these processes, and each technology has a characteristic sound.
What Are Your Symptoms?
We use compression for several reasons in sound reinforcement. One is to prevent the soft sounds produced by the instruments and vocals on stage from getting lost or masked by loud instruments, enthusiastic crowds or poor room acoustics. As an example, a singer may have power in certain areas of their vocal range but not in others. Or perhaps there are changes in volume when a guitar player switches between clean, crunch and lead sounds, making it difficult to mix. By evening out these differences, compression improves intelligibility and reduces (note that I did not say eliminates) the amount you’ll have to ride the fader on certain instruments.
Compression can also be used to protect our gear. Setting a compressor’s threshold to the right level can allow a P.A. to get loud, yet prevent an overzealous engineer from damaging amps and speakers by overdriving them. Restaurants or bar P.A.s may use compression to keep the level of background music consistent, even when the program material varies. And, of course, we can use compression as a special effect. (For a closer look at the parameters you can tweak, see sidebar below.)
Working with Compression
There are no rules regarding which instruments can or cannot be compressed. Regardless of genre, sounds that typically get compressed include vocals, bass, kick and snare, followed by keyboards and guitars. A compressor is usually patched on the channel insert, thus dedicating it to one particular signal. This is generally a good thing as it enables you to optimize settings for that particular channel. (You could group several channels and insert one compressor on the group, but that doesn’t work very well due to the interaction between the compressor and the various sounds). The tradeoff when using an analog desk with hardware compressors is that you’ll need a separate compressor for every channel you wish to process. Digital or software-based consoles usually provide a compressor on every input (and often output) channel, or you may be able to use many “instances” of a plug-in.
Whether analog or digital, we need to pay attention to where in the signal path the compressor is patched: it really needs to be pre-fader. Fortunately, the channel inserts on most hardware consoles are in fact, pre-fader. A software-based mixing system may provide the option to place the compressor pre- or post-fader, and that can be a critical consideration. If the compressor is inserted post-fader, every time you move the channel fader, the action of the compressor will change. As you ride the fader up, you’ll get more compression, and as you move the fader down, you’ll get less. Not good. When the compressor is patched pre-fader, you can set a threshold that is strictly related to the sound arriving at the microphone, not related to your manipulation of the fader.
Some hardware compressors allow you to switch the operating level between +4 and -10 dB; you need to make sure the unit matches the operating level of your console’s inserts or they won’t play nicely in the sandbox. You’ll probably need to consult the manual, but if it’s a TRS insert (for example, tip = send, ring = return) it’s most likely running unbalanced at -10. If you set the compressor to -10 and the console insert operates at +4, you’ll get too much compression (even with the threshold way up) and you’ll find that the input of the comp overloads easily. If the comp is at +4 and the console insert is at -10, you’ll need to bring the threshold way down to get any compression and you’ll have to crank the output gain, adding noise.
Next month we’ll look at specific compression settings for various instruments.
Examine Your Parameters
Compressors typically provide control over several parameters (adjustments). Here are some of the parameters you can tweak to control what the audience hears:
Threshold is like a sensitivity control. It is the signal level above which compression occurs. The higher the threshold, the less often the compressor acts and vice versa. You can think of threshold like setting the thermostat on your air conditioner: when the temperature reaches a certain point, the A/C kicks on and lowers the temperature. Some compressors reduce gain gradually as the signal level approaches threshold, making the effect subtle (see “knee,” below). Not every compressor has a threshold control. For example, in the classic UREI/UA 1176 (and its emulations), threshold is set internally based upon what ratio has been chosen.
Ratio is the difference between the input level and output level. As an example, let’s say the ratio is set 4:1. If the input is above threshold and rises 8 dB, the output will go up only 2 dB. If we changed that ratio to 8:1, the output would increase only 1 dB. When signal level is below threshold, ratio is essentially meaningless (there are some exceptions).
Attack time is the amount of time it takes for compression to start, once the signal level rises above threshold. Attack time is usually expressed in milliseconds, sometimes microseconds. With certain compressors, setting a fast attack time can produce more compression than if the attack time was slow.
Release Time (sometimes referred to as recovery) is the amount of time it takes for the compressor to “let go” once the audio gets quiet and the signal level drops below the threshold. This is important because when an instrument returns to a soft level, we no longer need to compress it. Release time determines how long the compressor holds the level down and is expressed in a range from milliseconds to seconds.
Knee: Threshold is not really a clear line in the sand. Knee refers to the way a compressor behaves as signal approaches threshold. A “hard-knee” compressor will reduce gain only when the signal level actually reaches the threshold, but a “soft-knee” allows the compressor to gradually begin acting when the signal is a few dB below threshold. Hard-knee compression can be quite audible, whereas soft-knee makes the onset of compression difficult to detect. dbx’s OverEasy is a form of soft-knee compression. Knee is illustrated in Fig. 1.
Gain or Output Level (sometimes called make-up gain): Compression, by definition, is a form of gain reduction and can make the signal level much lower than it was originally. This control allows you to compensate for lost gain.
Limiting: A limiter is essentially a compressor with a high ratio and a very fast attack time. A ratio of 10:1 or higher is generally considered limiting and is useful when you want to stop signal level from getting any louder once it reaches a certain point. Limiting can be used to govern the input to a power amp, thus protecting our amps and speakers. Limiters are also useful when combined with a compressor for taming very dynamic singers (or singers who have really bad microphone technique). The limiter puts a stop on the loudest parts of the performance while the compressor smooths minor variations in volume.
Metering: Ideally, a compressor provides three meters: one each for input and output level, and a third to show gain reduction. Unfortunately, meters cost money so on a hardware compressor, it’s more common to find one meter that can be switched between the three, or one that switches between input and output level and perhaps a second meter for gain reduction. Gain reduction meters usually work backward. A VU meter will sit at 0, indicating no gain reduction, while an LED meter will light progressively from right to left as gain reduction increases.
